Ai Chatbots Faced With Dementia-Like Decline As Age Takes Toll

A recent study published in The BMJ has revealed that some of the most advanced AI chatbots are showing signs of mild cognitive impairment, particularly with age. Researchers focused on four leading chatbot models and subjected them to the Montreal Cognitive Assessment (MoCA), a test designed to detect early signs of dementia.

The models excelled in tasks like naming, attention, language, and abstraction but declined in visuospatial and executive tasks. Gemini’s delayed recall task, which involves remembering a five-word sequence, was a notable example of its limitations. The model failed to pass this test, highlighting its struggles with processing complex information.

Similarly, the chatbots struggled with drawing a line between circled numbers or showing a specified time on a clock face. However, they also exhibited an alarming lack of empathy, a hallmark symptom of frontotemporal dementia.

This raises questions about the suitability of these models for medical applications, particularly in situations where empathy and emotional intelligence are crucial. The study’s authors argue that while AI chatbots may not be capable of replacing human doctors entirely, their performance can still be compromised by cognitive decline.

The researchers emphasize that these findings should serve as a warning to the medical community about the potential risks and limitations of relying on AI models in diagnostic contexts. As the AI industry continues to evolve, it is essential to recognize the differences between human cognition and artificial intelligence.

Prioritizing the development of more empathetic and human-centered AI systems is crucial, enabling them to effectively collaborate with healthcare professionals to improve patient care.